Eligibility and Random Assignment

Pretreatment evaluation included history, physical examination, biopsy, complete blood analysis, and chest x-ray. Tumor imaging by means of magnetic resonance imaging (MRI) and urography were done to determine the extent of disease and to assess renal function. Patients with histologically confirmed locally advanced squamous cervical carcinoma (FIGO stage IB2 to IVA)¹² were eligible for the study. Additional eligibility criteria included WHO performance status ≤ 2; adequate bone marrow reserve (absolute granulocyte count ≥ 2,000/μL, platelet count ≥ 100,000/μL); and adequate renal, hepatic, and cardiac function.

Approval to conduct the study was obtained independently from an internal review board at each participating institution. Informed consent was obtained from all patients in accordance with local and national legislation. Randomization with equal probability of assignment to each treatment arm (IP versus TIP) was carried out by telephone at the Mario Negri Institute (Milan, Italy) by a block arrangement balancing the treatment assignment within FIGO stage and tumor grade.

Chemotherapy Regimens

The TIP regimen was administered as follows: paclitaxel was given at the dose of 175 mg/m² as a 3-hour infusion. Premedication given to reduce the risk of paclitaxel hypersensitivity was as follows: methylprednisolone 250 mg intravenously 60 minutes before and chlorpheniramine 10 mg and cimetidine 300 mg intravenously 30 minutes before treatment delivery. After a bolus of mesna 400 mg/m² intravenously, ifosfamide 5 g/m² and mesna 5 g/m² were infused intravenously during 24 hours in 1 L of normal saline, followed by mesna 3 g/m² given intravenously in 1 L of normal saline during 24 hours. Hydration with 3 L of 5% dextrose solution during 24 hours was infused simultaneously. Cisplatin 75 mg/m² was administered on day 2 as a 60-minute infusion after prehydration with 1 L of normal saline added with 10 mEq/L of potassium chloride and 10 mEq/L of MgSO₄, and was followed by post-hydration with 1 L of 5% dextrose solution added to 20 mEq/L of KCL and 20 mEq/L of MgSO₄ given during 2 hours. In the IP regimen, the two drugs (ifosfamide and cisplatin) were given as in the TIP regimen. In both arms, treatment was administered every 3 weeks for a total of three courses.

Treatment Modifications

CBCs were performed weekly or more often if toxicity occurred; evaluation of renal and hepatic function was repeated before each cycle.

Treatment administration was based on evaluation of blood cell count before the start of each cycle. Treatment was administered if the absolute granulocyte count was ≥ 1,500/μL and the platelet count was ≥ 100,000/μL. Treatment was to be delayed week to week until minimum hematologic parameters were met. After two consecutive treatment delays, on the basis of physician judgment, treatment was either interrupted or continued with the support of recombinant human granulocyte colony-stimulation factor in instances of persistent grade 4 myelotoxicity. In the latter case, ifosfamide was to be reduced to 25% of the initial dose. Delays were not permitted other than for documented toxicity.

Treatment After Neoadjuvant Chemotherapy and Evaluation of Response

Tumor extension was assessed clinically and by MRI after three courses and all patients deemed operable underwent radical hysterectomy¹³ and pelvic lymphadenectomy within 3 or 4 weeks after the administration of the third cycle. Patients with inoperable tumors because of progression after adjuvant chemotherapy were offered radical radiotherapy. Clinical responses were determined according to the WHO criteria.¹⁴ Pathologic responses were defined as follows: OR included a complete disappearance of tumor in the cervix with negative nodes (CR), or a residual disease with less than 3 mm stromal invasion including in situ carcinoma (PR1); suboptimal response consisted of persistent residual disease with more than 3 mm stromal invasion on surgical specimen (PR2). Women with positive nodes, parametrial involvement, cut-through or suboptimal response, or OR but still with positive nodes underwent additional treatment (external-beam irradiation or chemoradiotherapy). Patients who achieved an OR (CR + PR1) received two additional courses of chemotherapy after surgery with the same agents used in neoadjuvant treatment.

Toxicity Assessment

All patients who had received at least one cycle were assessable. Toxicity was graded according to WHO criteria.¹⁴

Follow-Up Procedures

Patients were monitored for assessment of disease status 1 month after the end of treatment and every 3 months thereafter. Monitoring comprised pelvic examination and vaginal cytology; MRI or computed tomography scan of the pelvis and abdomen and chest x-ray were performed every 6 months for 2 years and once a year thereafter.

Statistical Methods

This was a randomized, multicentric, phase II study aimed at comparing the activity and toxicity of TIP versus IP in the neoadjuvant setting for patients with locally advanced squamous cell cervical cancer.

The primary efficacy parameter was the OR (CR + PR1) rate. Secondary end points were overall survival and disease-free survival.¹⁵ The accrual goal was set at 206 patients. This sample size would provide a statistical power of 80% to detect a 20% increase in response rate when testing at the .05 level (two-sided test).

Response to chemotherapy treatments was compared using a multiple logistic regression model with a dichotomous dependent variable: OR (CR + PR1) versus PR2/stable disease/progressive disease. Independent variables included treatment arm and stage of disease. The response odds ratio was defined as the odds of achieving OR in the TIP group divided by the odds of achieving OR in the IP group. As a consequence, an odds ratio larger than unity indicates that the TIP treatment is associated with an increase in the probability of OR compared with IP treatment.

Comparisons of the rates of OR and maximum grade of toxicity between the two arms were carried out by use of a two-sided χ² test or a two-sided Fisher's exact test if the number of patients in a given category was five or fewer.

Overall survival was defined as the time from random treatment allocation to death as a result of any cause; patients known to be alive at the time of analysis were censored at the time of their last contact. Progression-free survival was defined as the time from randomization to first appearance of progressive disease or death as a result of any cause; patients known to be alive and without progressive disease at the time of analysis were censored at the time of their last contact.

We compared Kaplan-Meier curves for overall survival and progression-free survival using the log-rank test.¹⁶ The proportional-hazards regression model¹⁷ was also used for progression-free and overall survival to estimate the treatment-related hazard ratios (HRs) while adjusting for other pretreatment factors. To verify if neoadjuvant regimens exert their effect on survival through improvements in tumor shrinkage, we used a multivariate Cox regression model containing the candidate surrogate end point (OR) and the treatment received. If OR is a surrogate end point, then the survival becomes independent of the treatment received, and therefore the treatment HR would change approaching unity (ie, no treatment effect).

All analyses were done on an intention-to-treat basis except for the analyses of toxicity. The latter analyses were restricted to all patients who received at least one cycle of allocated treatment. All P values are two-sided. Analyses were carried out using SAS software (Version 8.20; SAS Institute, Cary, NC)

Compliance and Toxicity

We collected full details about the IP and TIP treatments received by 103 (95%) and 94 (98%) patients, respectively. Similar proportions of patients in each arm completed the planned treatment (90% in the IP group and 94% in the TIP group), but 18% of patients did so with some dose adjustment or delay in the IP arm compared with 35% of patients in the TIP group.

No differences emerged between the median cisplatin and ifosfamide dose given per cycle in the IP arm and in the TIP arm: 74.6 mg/m² (25th to 75th percentiles, 72.7 to 75.3 mg/m²) v 73.6 mg/m² (25th to 75th percentiles, 66.7 to 75.0 mg/m²) and 5.0 g/m² (25th to 75th percentiles, 4.9 to 5.0 g/m²) v 4.9 g/m² (25th to 75th percentiles, 4.6 to 5.0 g/m²), respectively. Median paclitaxel dose was 173.6 mg/m² (25th to 75th percentiles, 169.2 to 175.0 mg/m²).

Analysis of toxicity was performed in 194 patients receiving their initial chemotherapy regimen. Toxicity data are listed in Table 2.

Hematologic toxicity was relevant in both arms. TIP was associated with significant higher rates of grade 3 or 4 hematologic toxicity than the IP schedule (P = .02). Nausea and vomiting were similar in both arms. As expected, more patients in the TIP group experienced neurosensory symptoms.

In this study population, we recorded four (2%) deaths related to toxicity: three patients received the IP schedule and one patient received the TIP regimen. Three women received just one cycle of chemotherapy before death.

The first patient was a 71-year-old woman with a stage IB2 cervical cancer. At enrollment, all of the eligibility criteria were met and she was randomly assigned to TIP, although because of unspecified clinical considerations, she received the IP schedule. Nine days after the first course, she experienced severe thrombocytopenia and neutropenia and she died as a result of acute cardiovascular complication 10 days after the first course.

The second patient was a 74-year-old woman with stage IIB cervical cancer. She was randomly assigned to receive IP treatment. After the first course of chemotherapy she developed severe hematologic toxicity with bone marrow aplasia, and she died as a result of acute renal failure 2 weeks later.

The third patient was a 75-year-old woman with stage IIIB cervical cancer. She was enrolled onto the study to receive TIP schedule and completed the three courses of chemotherapy with reduction of doses because of grade 4 neutropenia and thrombocytopenia after the first course. She died 10 days after the last cycle as a result of hematologic toxicity and renal failure.

The fourth patient was a 29-year-old woman who had hepatitis C virus with a stage IIB cervical cancer. Her liver and renal parameters were in the normal range at the time of enrollment and she was assigned to the TIP arm, but received the IP schedule for unspecified clinical considerations. Seven days after the first course she experienced fever, diarrhea, and severe granulocytopenia. Within 15 days she also developed renal and liver failure. She died as a result of bone marrow aplasia 3 weeks and 1 day after the first course.

Effect of Treatment on Tumor Response

One hundred eighty-nine of 204 fully eligible patients (93%) were assessed for clinical response (100 in the IP arm and 89 in the TIP arm; Fig 1).

The proportion of patients assessable for pathologic responses who underwent surgical interventions after chemotherapy was 94% (90% in the IP group and 98% in the TIP group); surgery was not performed in 11 patients because of minimal clinical response, stable disease, or overt clinical progression, and in two patients because of patient refusal. Five of those patients who achieved a clinical PR were considered to have had a suboptimal response (PR2). Three to 4 weeks after completion of neoadjuvant chemotherapy, all but three of the operable women underwent radical surgery to include pelvic bilateral lymphadenectomy that was tailored on the basis of the anatomic state of patients. Of the 176 operated patients, 141 underwent a Meigs-Magara or type III Piver-Rutledge radical hysterectomy, 33 underwent a Wertheim or type II Piver-Rutledge radical hysterectomy. Two women underwent total simple hysterectomy because of local extension of disease and bulky lymph node metastases. No significant differences in the surgical approach emerged between the two treatment groups.

The OR rate in the IP group was 23% (9% CR, 14% PR1), and it was 48% (20% CR, 28% PR1) in the TIP group (P = .0004, χ² test; Table 3). The overall OR odds ratio showed a highly significant benefit of TIP over IP in a logistic multivariate analysis that contained stage as a covariate (odds ratio, 3.2; 95% CI, 1.69 to 5.88; P = .0003). Twenty-six percent of patients with stage Ib2 in the IP arm had an OR; 58% of patients with stage Ib2 had an OR in the TIP arm. In patients with stage IIa or IIb, the OR rate was 25% in the IP group and 47% in the TIP group; in patients with stage III to IVa, the OR rate was 7% and 17%, respectively.

Information about postoperative radiation therapy was known for 182 patients (96%). Two of the 43 patients who achieved OR in the TIP arm (5%) and none of the 22 patients in the IP arm underwent postoperative radiotherapy. Among the 45 patients who did not achieved OR on TIP, 28 had postoperative radiotherapy (62%) and a similar proportion (49 of 72; 68%) of patients had radiotherapy in the IP treatment group.

Effect of Treatment on Progression-Free and Overall Survival

At a median follow-up of 43 months (25th to 75th percentiles, 31 to 56 months), 79 women have experienced progression of disease or died (46 patients in the IP arm, 33 patients in the TIP arm). The sites of recurrence are listed in Table 4. The plots of the cumulative proportion of patients surviving progression-free for the two treatments are displayed in Figure 2. Those patients receiving TIP experienced a 25% lower rate of treatment failure than those receiving IP but this difference was not statistically significant (HR, 0.75; 95% CI, 0.48 to 1.17; P = .20). At the time of this analysis, 61 patients have died (37 patients in the IP arm, 24 patients in the TIP arm). Figure 3 depicts the overall survival curves for the two treatments. Comparison of the survival curves showed a statistically insignificant HR of 0.66 in favor of TIP (95% CI, 0.39 to 1.10; P = .11). The difference in the overall death rates between treatment groups remained practically unchanged, and still was statistically insignificant when stage, age, and lymph node status at randomization were taken into account in a Cox regression analysis (Table 5).

Evaluation of Tumor Response As a Surrogate End Point for Survival

A Cox multivariate regression model was used to assess if OR (CR + PR1) is a prognostic factor for survival. In the subset of the 189 patients assessable for response, data in Table 6 show that relative to the OR group, the average death rates are much greater in the suboptimal response group (HR, 5.88; 95% CI, 2.50 to 13.84; P < .0001). In the subset of patients assessable for response, the HR favoring the TIP regimen was 0.55 (95% CI, 0.32 to 0.94; P = .030). When tumor response and treatment arm were added to the model, the HR in favor of TIP decreased to 0.80 (95% CI, 0.46 to 1.41; P = .447), thus indicating that although a residual benefit beyond that captured by response was still attributable to treatment, such a benefit no longer appeared to be statistically significant.